I have a computer with a Ryzen 5 2600 and an old Celeron 2955U chromebox both running linux, and right now the chromebox is getting more credit per task than the newer, more powerful server. How can this be?

Up until this week, the Ryzen machine was getting between 300 and 400 credits per task. I noticed the drastic drop in credit seems to correspond to jobs with graft in their name. Is this some sort of credit calculation bug, or am I wasting electricity running these jobs on the Ryzen machine?

Ryzen 5 - https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=5426050
cd28_1yjd_graft_v1_SAVE_ALL_OUT_IGNORE_THE_REST_9pj2td8o_1013365_1_0
https://boinc.bakerlab.org/rosetta/result.php?resultid=1269585997
Credit: 67.46

======================================================
DONE ::  3087 starting structures  28809.2 cpu seconds
This process generated   3087 decoys from    3087 attempts
======================================================
BOINC :: WS_max 9.68561e+08
20:00:15 (1002): called boinc_finish(0)

Celeron - https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=4337761
cd28_1yjd_graft_v1_SAVE_ALL_OUT_IGNORE_THE_REST_7ak2hr4u_1013377_1_0
https://boinc.bakerlab.org/rosetta/result.php?resultid=1269371948
Credit: 82.24

======================================================
DONE ::   899 starting structures  28821.2 cpu seconds
This process generated    899 decoys from     899 attempts
======================================================
BOINC :: WS_max 9.91904e+08
16:20:14 (2662): called boinc_finish(0)

I have a computer with a Ryzen 5 2600 and an old Celeron 2955U chromebox both running linux, and right now the chromebox is getting more credit per task than the newer, more powerful server. How can this be?

The benchmarks for the Ryzen haven't been run, so the default values are there. Eventually allocated Credit will improve, but if you run the benchmarks it will speed the process up, and stop the low Credit for new work types when the next new work type comes out.

---

Grant
Darwin NT

I have a computer with a Ryzen 5 2600 and an old Celeron 2955U chromebox both running linux, and right now the chromebox is getting more credit per task than the newer, more powerful server. How can this be?

The benchmarks for the Ryzen haven't been run, so the default values are there. Eventually allocated Credit will improve, but if you run the benchmarks it will speed the process up, and stop the low Credit for new work types when the next new work type comes out.

So should EVERYONE with a Ryzien cpu run the benchmark program? Or just certain Ryzien cpu's? And where do you find the benchmark program to run it?

He's referring to the BOINC Benchmark.

---

The benchmarks for the Ryzen haven't been run, so the default values are there. Eventually allocated Credit will improve, but if you run the benchmarks it will speed the process up, and stop the low Credit for new work types when the next new work type comes out.

Thanks Grant! The Ryzen machine is pretty new to BOINC, so it makes sense that the benchmarks haven't been updated from the default values. I didn't know that those stats were used to calculate credit. At least now I know there isn't something wrong with my computer. It's the first one I built myself, and it's a big step up from a chromebox.

Thanks Grant! The Ryzen machine is pretty new to BOINC, so it makes sense that the benchmarks haven't been updated from the default values. I didn't know that those stats were used to calculate credit. At least now I know there isn't something wrong with my computer. It's the first one I built myself, and it's a big step up from a chromebox.

That's looking much better.

Measured floating point speed  5315.09 million ops/sec
       Measured integer speed 91514.58 million ops/sec

The default value is 1000.

---

Grant
Darwin NT

He's referring to the BOINC Benchmark.

Yep. Advanced view, Tools, Run CPU benchmarks.
it should run when BOINC is first installed, but if it isn't then the default value is used and initial Credit for any new type of work done will be extremely low on any reasonably recent CPU.

On any system BOINC has just been installed on (and particularly on a system that's already got BOINC on it but the CPU has just been upgraded) it's worth manually running it just to be sure. As long as it's done when there's nothing else working the system hard, it shouldn't need doing again (at least until the next CPU upgrade happens).

---

Grant
Darwin NT

He's referring to the BOINC Benchmark.

Yep. Advanced view, Tools, Run CPU benchmarks.
it should run when BOINC is first installed, but if it isn't then the default value is used and initial Credit for any new type of work done will be extremely low on any reasonably recent CPU.

On any system BOINC has just been installed on (and particularly on a system that's already got BOINC on it but the CPU has just been upgraded) it's worth manually running it just to be sure. As long as it's done when there's nothing else working the system hard, it shouldn't need doing again (at least until the next CPU upgrade happens).

Okay thanks, my machines have been around for a long time so no need to do that for me, they pretty much are what they are, most are too old to do a worthwhile upgrade anyway. A few AMD ones could probably go from 6 to 8 cores but that's not emough for me to worry about since they are old enough to be replaced when I get the next pc's, I'm looking at 16+ core pc's now and with Christmas coming the prices could drop down to where my wife won't complain if I get one new pc and replace 2 older ones.